Content
68%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A highly actionable playbook whose copy-paste system prompt template and worked examples are its main strength. Its weaknesses are redundancy — freshness and citation rules are repeated three times across sections and the final worked example restates completion criteria — and the absence of an explicit, validated step sequence for the authoring workflow itself.
Suggestions
State the freshness/citation rules once inside the system prompt template and reduce the 'Freshness policy' and 'Required behavioral rules' sections to a short enforcement checklist, cutting the repeated statements of the same rules.
Add an explicit numbered authoring workflow (pick identity → check available capabilities → fill template → verify) ending with a verification step that checks the produced prompt against the completion criteria.
Explain or remove the undefined 'from the form snapshot' reference in the worked example's model-selection line, since no form or model snapshot is defined anywhere in the skill.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The 60-line system prompt template is the core deliverable and earns its place, but the freshness/citation rules are stated three times ('Freshness policy' section, the template's 'Freshness and source policy', and 'Required behavioral rules'), and the closing worked example restates the completion criteria again. Roughly a quarter of the body is redundant restatement, fitting 'mostly efficient but includes some unnecessary explanation or could be tightened' rather than the minor-trim 4 anchor. | 3 / 5 |
Actionability | The body provides a complete, fill-in-the-placeholder system prompt template with an embedded worked example, concrete identity naming patterns, a prioritized capability list, and an end-to-end worked example ('Build me an agent that researches AI startups' → named agent with description, model, tools, and prompt excerpt). For an instruction-only skill this is copy-paste-ready guidance covering the common case, matching the 5 anchor; it is not the 4 anchor because there are no material gaps in the guidance (the undefined 'form snapshot' phrase is peripheral to the main flow). | 5 / 5 |
Workflow Clarity | The components a builder needs are all present and ordered sensibly (identity → freshness policy → template → enforcement rules → capabilities → anti-patterns → example), but the authoring process itself is never sequenced as steps and has no explicit validation checkpoint for the produced prompt — the 'enforce these rules' checklist is implied rather than a verify step. This matches 'steps listed but validation gaps; checkpoints missing or implicit' rather than the 4 anchor's mostly-present checkpoints. | 3 / 5 |
Progressive Disclosure | The body has well-organized sections and no buried or nested references, and the prompt template legitimately belongs inline. However at ~130 lines with no bundle files, the full template and worked example could live in a references/ file to keep SKILL.md lean, a minor organization gap consistent with the 4 anchor; it is not the 5 anchor, which requires content appropriately split across well-signaled files (and the >50-line simple-skill exception does not apply). | 4 / 5 |
Total | 15 / 20 Passed |